Incentive salience in human cognition during health and disorder

Project Information

INSENSE

Grant agreement ID: 804360

DOI

10.3030/804360

Project closed

EC signature date 22 July 2019

Start date 1 August 2019

End date 31 July 2025

Funded under

• EXCELLENT SCIENCE - European Research Council (ERC)

Total cost

€ 1 458 033,00

EU contribution

€ 1 458 033,00

Coordinated by

THE UNIVERSITY OF BIRMINGHAM
United Kingdom

Periodic Reporting for period 4 - INSENSE (Incentive salience in human cognition during health and disorder)

Reporting period: 2024-02-01 to 2025-07-31

Summary of the context and overall objectives of the project

Incentive salience is a form of motivation for reward that is triggered by environmental cues. These come to be ‘wanted’: they create an urge or craving for approach and consumption that influences choice and guides action. Stimuli imbued with incentive salience are thought to become salient, attention-drawing, and impossible to ignore, and a leading theory of addiction proposes that drug stimulation of the brain’s reward system may create intense and abnormal incentive salience for drug-related stimuli. Consistent with this, work with animals has linked incentive salience to signaling in mesocorticolimbic brain systems, and the release of nigrostriatal dopamine in particular. But direct investigation of incentive salience in human cognition is sparse, and the application of ideas from animal research to our understanding of human incentive salience has led to pervasive ambiguity and misunderstanding. Direct research with humans is needed to determine the role that incentive salience plays in normal and abnormal human cognition, and particularly in addiction. The objective of INSENSE was therefore to use cutting-edge tools from cognitive neuroscience to a.) characterize the computational and neural substrates of human incentive salience, and b.) determine how failures in these systems underlie addictive human behaviour. This was accomplished through the combined use of techniques like psychopharmacology, electroencephalogram, multivariate pattern analysis of functional magnetic resonance data, and computational modelling in order to index, characterize, and manipulate the neural representation of naturalistic reward-associated stimuli.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

From the start of the project, we investigated how visual attention and learning mechanisms assign priority to objects in complex environments, with a particular focus on incentive salience and its neural implementation. Early work established that visual attention is structured around objects, demonstrating with EEG that task-irrelevant object boundaries systematically shape anticipatory attentional deployment in human visual cortex, resolving longstanding ambiguity in behavioural accounts of object-based selection. This provided a mechanistic foundation for understanding how visual systems define candidate units for learning and value assignment.

Subsequent phases focused on how incentive salience is acquired, expressed, and suppressed for real-world object categories. Using machine-learning analysis of EEG and event-related potentials during naturalistic visual search, we showed that objects associated with reward rapidly acquire incentive salience, altering early sensory encoding. Critically, however, humans can also exert fast, stimulus-triggered suppressive control over incentive salience when reward-associated objects are known to be irrelevant. This suppression degrades neural representations of reward-associated distractors within ~200 ms, preventing attentional capture while preserving behavioural goal performance.

Together, these results demonstrate that incentive salience operates directly on visual representations and is dynamically regulated by cognitive control. This reframes reinforcement learning in vision as an object-centred process in which learned value reshapes perceptual encoding itself, rather than acting solely at post-perceptual decision stages. The findings imply that visual reinforcement learning involves continuous negotiation between value-driven incentive salience and top-down suppression.

Clinically, this has direct implications for addiction research. Maladaptive incentive salience attributed to drug-associated cues is a core feature of substance-use disorders, driving craving, attentional bias, and relapse. Our results identify rapid neural suppression of incentive salience as a candidate protective mechanism and a potential treatment target. By characterising when and how incentive salience can be overridden in visual cortex, this work provides a principled framework for developing biomarkers of vulnerability and for designing interventions that strengthen control over pathological incentive salience in addiction and related compulsive disorders.

Results have been disseminated in 20 journal articles, a variety of conference presentations and proceedings, and dozens of conference posters.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

Human incentive salience is currently thought to be a DA-driven Pavlovian mechanism that acts on the perceptual representations of environmental stimuli. It has been interpreted as a critical mechanism in the instantiation of human addictive behaviour. But this characterization is debatable. First, there is little direct evidence that DA is the prime mover in human incentive salience. It is very likely that other systems are also involved in instantiating the effect. For example, serotonin projections from the caudate tail may link craving to sensory input, and glutamatergic signaling in amygdala may bind valence to discrete stimuli. These mechanisms may operate independently of DA to create incentive salience, or they may rely on DA signaling to be initiated. The causative role of DA in human incentive salience is therefore directly tested in WP1 of the INSENSE project using combined psychopharmacology and neuroimaging. Second, it is unclear if incentive salience reflects model-free Pavlovian learning. It is clear that human choice behaviour does not, but this could reflect cognitive deliberation unrelated to incentive salience. To test the possibility that incentive salience can be accounted for by model-free learning, linking it to the broader literature on reinforcement learning, we must develop an index of the representation of reward-associated stimuli that can be modelled independently of overt response. WP4 proposes to do so using MVPA analysis of fMRI. Third, it is unclear that incentive salience reflects variance in the sensory or perceptual encoding of environmental stimuli, rather than effects at later cognitive stages like decision-making or motor control. WP2 addresses this issue using time-resolved neuroimaging and by manipulating sensory plasticity using transcranial electrical stimulation. Finally, the idea that human addicts are biased to encode information about drug-related stimuli has been tested primarily through observation of overt response. But overt response is sensitive to effects at cognitive stages other than sensation and perception. WP3 tests the idea that drug-related stimuli are perceptually and attentionally prioritized in addiction using eye-tracking and concurrent neuroimaging. The idea of incentive salience remains grounded in animal research, and this creates ambiguity and misunderstanding in its application to human cognition in health and disorder. The INSENSE project was designed to use cutting-edge tools from cognitive neuroscience to test the efficacy of this idea as an account of human behaviour and brain activity.

Figure 1 – Correlational MVPA analysis of fMRI data. The pattern of voxelwise activity elicited in v